Computer-Aided Detection in the Mammographic Detection of Breast Cancer: Where, Why, and How With Screen-Film, Computed Radiography, and Full-Field Digital Mammography

Abstract

Mammography remains the mainstay for screening for breast cancer. However, screen-film mammography (SFM) is an imperfect examination with 10% to 35% of breast cancers not mammographically visible. A recently developed and implemented approach to the improved diagnosis of breast cancer is computer-aided detection (CAD), where computer algorithms and neural networks are used to mark potential areas of abnormalities on the mammogram for the radiologist to evaluate and determine whether additional workup is indicated. CAD has been extensively studied retrospectively and prospectively and has demonstrated a 7% to 20% improvement in breast cancer detection. This manuscript will review the current literature of CAD, the principles of computer assessment of mammograms, and where the state of the art is going in this transitional time from SFM to full-field digital mammography (FFDM) as well as computed radiography for mammography. CAD implemented with mammography has demonstrated improvements in breast cancer detection for SFM, computed radiography, and FFDM. This manuscript will review the state of the art of CAD for the detection of breast cancer.